Optimized multimodal logistics planning of modular integrated construction using hybrid multi-agent and metamodeling

Multimodal logistics (ML), which involves multiple transportation modes, has been increasingly used in many Modular integrated Construction (MiC) projects. However, the literature lacks decision support systems (DSS) to simulate, analyze, and optimize ML in MiC (ML-MiC). This paper fills this gap by achieving the following objectives: 1) simulate the internal operations of ML-MiC stakeholders (e.g., manufacturers, logistics service providers, contractors) and their interactions; 2) identify the significant decisions that impact the key performance measures (KPMs) of ML-MiC; and 3) obtain the near-optimum decisions that improve the sustainability of ML-MiC. These objectives are achieved by developing a holistic modelling approach that integrates three methods. First, hybrid multi-agent simulation models the communications between ML-MiC stakeholders and their internal operations. Second, design of experiments (DOE) reveals the main and interaction effects between logistics and construction decisions that significantly affect KPMs, such as the project duration, total costs, and carbon emissions. Third, metamodeling finds the near-optimum logistics and construction decisions (e.g., trucks' number, their dispatching time, ship capacity, inventory, resource planning) that enhance KPMs. The developed approach is applied to a real case study. The DOE analysis indicates that some logistics decisions significantly influence construction KPMs (e.g., project duration, construction costs, construction emissions) and vice versa, calling for more collaboration between stakeholders. Also, the optimized solutions reduce the project duration, total costs, and emissions by 28%, 50%, and 17%, respectively. This paper contributes by integrating three methods to model ML-MiC and enable its stakeholders to discern the impact of their decisions on multiple KPMs and optimize them toward more sustainable MiC. Given this paper's findings, future researchers are urged to investigate the success factors and barriers to applying ML in MiC. Also, the paper emphasizes the need to develop DSS that achieve a win-win collaboration and enhance communication between ML-MiC stakeholders. • Multimodal logistics (ML) are used to import modular units from overseas factories. • The literature lacks models to assess, analyze and optimize ML-related decisions. • A developed multi-agent model evaluates multiple project performance measures (PMs). • A diagnostic analysis reveals the significant ML decisions that impact each of PMs. • Metamodeling finds the optimum ML decisions to improve PMs of modular construction.